Summer 2008 RLM 11.176 MTWTHF 10:00 - 11:30 a.m. July 14, 2008 - August 18, 2008
fee: $33.99
CONSENT OF INSTRUCTOR MUST BE OBTAINED. COURSE NUMBER MAY BE REPEATED FOR CREDIT WHEN THE TOPICS VARY. SAME AS CAM 394C.
Course Description. A course in modern computationally-intensive statistical methods including simulation, optimization methods, Monte Carlo integration, maximum likelihood / EM parameter estimation, Markov chain Monte Carlo methods, resampling methods, non-parametric density estimation. Prerequisite: Graduate Standing and Mathematics 362K and 378K, or consent of instructor.
We will use the statistics package R and Mathematica. It is not necessary for students to purchase Mathematica because students can use it in the computer labs in the math department. The program R is a free program so I expect that all students will download it to use.
I will provide explanations of how to use these programs.
The grading in the course will be based on homework assignments and two take-home exams. Students may collaborate on the homework assignments, but the take-home exams must be done individually.
Textbook: Computational Statistics by Geof H. Givens and Jennifer A.Hoeting, Wiley, 2005. Textbook website with a link to Amazon and Wiley with purchase information. The textbook will be available in the PMA (RLM) library on reserve for this course (2-hour reserve) beginning about July 9. The textbook website above has the table of contents and Preface as well as other useful information.
I will make various additional materials available for the coure through Blackboard. Several other relevant books will also be available on reserve in the library.
If you read the Preface of the textbook, you will see that the authors assume familiarity with several statistical techniques that are beyond the level of the prerequisite for this course. We will not cover chapters that use all of these, and for those we do cover, I expect to teach those statistical topics as well as the computational topics. In particular, students should already be familiar with maximum likelihood estimation and regression. Other topics will be introduced in the course as needed.
There is a suggestion in the Preface for the chapters/topics to be covered in a one-semester course. This seems too ambitious to me for this particular class, so I believe we will follow a "more leisurely pace" as they say in the Preface. It will be a small class and it will be a discussion / lecture class more than mainly a lecture class.